DNA-based assays for the detection of pathogenic fungi in clinical samples are now beginning to be marketed in some countries. This section will not attempt to list exhaustively all studies on this topic, but will sample the current published literature. Brief descriptions of the methods can be found in the section on Fungal Identification. Assays have been described for the detection of fungal DNA from nearly all body sites, but particularly blood (serum, plasma, and whole blood), respiratory fluids, CSF, ophthalmic materials, and dermatologic samples. Most molecular diagnostic assays are PCR-based, to take advantage of the increase in sensitivity offered by the many-fold amplification of PCR targets as well as the specificity offered through appropriate primer/
probe design. The LightCycler SeptiFast (Roche Molecular Systems) was the first commercial in vitro diagnostic assay for PCR-based detection of fungal pathogens in whole blood [71]. The MYC Assay (Myconostica, Manchester, UK) is currently undergoing evaluation as a molecular beacon assay to detect Aspergillus and Pneumocystis in respiratory sam- ples (www.myconostica.co.uk).
The most active area of research is in the development of assays to detect and identify Candida and Aspergillus spe- cies in blood. Generally, these tests are intended for prospec- tive monitoring of immunosuppressed patients at risk for fungal diseases, particularly hepatosplenic candidiasis or invasive forms of aspergillosis, and for the early detection of these difficult-to-diagnose diseases so that appropriate ther- apy can be initiated.
A recent review describes platforms and strategies for detection of Aspergillus DNA in blood [72]. In one of the earliest papers on this topic, Einsele et al. described a PCR designed to amplify a fungal-specific 18S ribosomal sequence from blood and then to identify the fungal pathogen using species-specific hybridization [73]. Since then, numerous papers have been published on this subject, using Southern blotting, PCR-EIA, nested PCR, and real-time PCR formats to detect Aspergillus DNA in bronchial lavage fluids, whole blood, plasma, or serum samples. A meta-analysis of 16 studies describing the use of PCR tests for the diagnosis of invasive aspergillosis was recently published [74], and results from more than 10,000 blood, serum, or plasma samples from 1,618 patients at risk were evaluated. These authors concluded that a single PCR-negative test is sufficient to exclude a diagnosis of proven or probable invasive aspergil- losis according to the European Organization for Research and Treatment of Cancer-Mycoses Study Group (EORTC- MSG) definitions; however, two positive tests are required to confirm the diagnosis due to a requirement for higher speci- ficity. They comment that future studies should distinguish between the use of the test to screen for the presence or absence of invasive aspergillosis in high-risk patients as
17 Laboratory Aspects of Medical Mycology
compared to use of the test to confirm the disease in symp- tomatic patients. The use of two different PCR targets might be appropriate for the latter purpose. They also found a need for standardizing the PCR methodology, including the nature of the sample, volume tested, DNA extraction methods, choice of target gene, detection of PCR products, and use of appropriate controls.
PCR detection of Candida species in blood has been investigated in many studies, due in part to the recognition that blood cultures are positive in fewer than half the cases of invasive candidiasis. Methods used have included PCR- RFLP, Southern blot hybridization, PCR-EIA, microarrays, LightCycler FRET, TaqMan, molecular beacons, biprobes and melt curve analysis, and Luminex [20]. Most studies have reported high sensitivity and specificity, although the fungal load in blood is low (as low as <1 cfu/mL in some cases). Several groups have reported that Candida DNA was detected more readily in serum and plasma than in whole blood, but there is no current consensus on the most appro- priate clinical sample for Candida diagnostics.
A number of publications describe the use of PCR to detect fungal pathogens in maxillary sinus tissue, nails, cor- neal tissue, vitreous fluids, and other ophthalmic samples [75]. In one representative study, nested PCR was used for detection and identification of C. albicans, A. fumigatus, and F. solani [76]. The authors noted that one patient sample was PCR positive but culture negative, suggesting that molecular testing may permit a diagnosis to be made even when organ- isms cannot be grown from the sample. A nested PCR has been described that detects DNA from Trichophyton and Microsporum species in human and veterinary dermatologic samples [77]. DNA could be detected even in the presence of contaminating bacterial organisms. DNA from various Malassezia species could be detected in dressings applied to skin lesions of affected patients [78].
Historically, diagnosis of fungal infections using fresh or paraffin-embedded tissue has been complicated by several problems. Fungi can be extremely difficult to distinguish from one another in tissue, contributing to misdiagnosis [6].
Many organisms demonstrate atypical morphology in tissue.
Many fungi, particularly mucormycetes (formerly zygomy- cetes) cannot be reliably cultured from tissue. Finally in many cases fungal culture is never ordered, so that histo- pathologic findings cannot be extended or confirmed. A number of PCR-based approaches and in situ hybridization assays have been developed to address these problems and to identify DNA recovered from tissue blocks. In the former category, a panfungal PCR targeting the ITS-1 region com- bined with DNA sequencing [79], a seminested PCR with sequencing [80], and PCR combined with Southern blot analysis [81] have been described. In the second category, in situ hybridization using probes directed against fungal ribo- somal DNA has been used for the identification of fungi in
tissue sections from five skin biopsies [82]. In situ hybridization is not helpful unless fungal organisms can be visualized using conventional histopathologic staining. Varying degrees of success have been reported in the recovery of DNA from formalin-fixed tissues. The formalin fixation step has been reported to cause degradation of amplifiable DNA as a result of DNA cross-linking. In cases where DNA cannot be recov- ered, immunohistochemical staining of tissue may be helpful (see Histopathologic Examination).
Molecular diagnostics offer great hope for the rapid detec- tion and identification of difficult-to-culture organisms, for detection of antifungal drug resistance, and for rapid diagno- sis directly from host tissues and fluids. At this time, many research laboratories offer “in-house” procedures for molec- ular identification of fungal isolates from culture plates, from tissue, or from body fluids. Their sensitivity, specificity, pre- dictive value, and clinical relevance have not always been rigorously investigated. It is to be hoped that in the future the relevance of these assays will be demonstrated and that they will be introduced to a broader audience in the clinical micro- biology laboratory community.
Antifungal Drug Susceptibility Testing
As with antibacterial compounds, tests designed to ascertain the minimum amount of drug needed to inhibit the growth of fungal strains in culture (minimum inhibitory concentration or MIC) are often assumed to be the most dependable means of determining the relative effectiveness of different antifun- gal agents and of detecting the development of drug-resistant strains. In addition, it is often assumed that the clinical out- come of treatment can be predicted from the results of in vitro testing of a patient’s isolate against a panel of potentially useful agents. Such an approach to the selection of antifun- gal agents has become more reasonable with the develop- ment of a reliable and reproducible reference procedure for in vitro testing of fungal species against antifungal agents and the demonstration of correlations with clinical outcome, at least for some forms of candidiasis. However, the pitfalls of assuming a correlation between the results of susceptibil- ity testing of other antifungal drugs and organisms in vitro and outcome in vivo should not be underestimated. With an increase in drug resistance demonstrated among such diverse fungi as C. glabrata, A. fumigatus, A. terreus, C. neoformans, and S. apiospermum [83–87], and the expanded range of antifungal agents, it is clear that the need for meaningful methods of in vitro testing of both new and established agents is increasing.
Starting in the mid-1980s there was a movement to estab- lish a standardized methodology for antifungal susceptibility testing, and to this end the CLSI (formerly the National
18 M.E. Brandt et al.
Committee for Clinical Laboratory Standards [NCCLS]) formed a Subcommittee on Antifungal Susceptibility Testing.
The first publication of a standardized methodology for anti- fungal susceptibility testing occurred in 1997 when the NCCLS published an approved reference method (docu- ment M27-A) for the in vitro testing of five antifungal agents (amphotericin B, flucytosine, fluconazole, itraconazole, and ketoconazole) against Candida species and C. neoformans by the broth macrodilution and microdilution methodolo- gies [88]. Although imperfect, this first document facilitated the establishment of interpretive breakpoints for fluconazole and itraconazole against Candida spp., specified a defined test medium (RPMI-1640 broth buffered to pH 7.0 with MOPS), and recommended an inoculum standardized by spectrophotometric reading to around 1,000 cells/mL. Two expanded editions of this document have now been pub- lished. The third document in this series is M27-A3 [89].
This document provides quality control (QC) limits at 24 and 48 h for amphotericin B, flucytosine, fluconazole, itra- conazole, voriconazole, posaconazole, ravuconazole, keto- conazole, anidulafungin, micafungin, and caspofungin against an expanded set of QC strains [90].
The M27-A reference series has permitted much greater standardization of the in vitro testing of antifungal agents;
however, several problems remain unresolved. These include the poor performance of the recommended culture medium in tests with some organisms and with amphotericin B, the method of end point determination, and the proper interpre- tation of trailing growth in tests with azole agents [91].
The defined culture medium described in the CLSI refer- ence procedure (RPMI-1640 broth buffered to pH 7.0 with MOPS) has proven less than ideal for the testing of some Candida species and C. neoformans. Increasing the glucose concentration of the medium from 0.2% to 2% results in bet- ter growth of most isolates of Candida species, making the visual determination of end points easier without signifi- cantly altering the observed MICs of amphotericin B, flucy- tosine, fluconazole, or ketoconazole [92, 93]. Another modification that appears to be helpful is the use of yeast nitrogen base medium in tests with C. neoformans [94].
Although not specifically recommended, both of these modi- fications have been noted in Supplement M27-S3 [95].
It has become evident that the use of the M27-A method to test Candida species against amphotericin B results in a restricted range of MICs. Given these results, there has been concern that the reference procedure might not detect resis- tance to amphotericin B. The difference in amphotericin B MICs between susceptible and resistant strains is more pro- nounced when antibiotic medium 3 (AM3) is used instead of standard RPMI-1640 medium [96]. However, complete sep- aration of resistant from susceptible isolates has not been achieved by using AM3 medium. In addition, AM3 is a non- standardized medium and lot-to-lot variation is a limiting
factor to its use [97]. It is recommended that laboratories that use this alternative medium introduce strains of Candida species with known amphotericin B MICs as controls.
The M27-A procedure relies upon the visual determina- tion of MIC end points. However, the recommended end points differ for different antifungal agents. For amphoteri- cin B, the end point is defined as the lowest concentration at which there is complete inhibition of growth. The end point for azoles and echinocandins for both macro- and microdi- lution testing has been defined as the point at which there is a prominent reduction in growth. For macrodilution testing, prominent reduction in growth has been shown to corre- spond to an 80% reduction in growth relative to that observed in the growth control. However, when the microdilution for- mat is utilized and read with a spectrophotometer, the speci- fied prominent visual reduction in growth best corresponds to a 50% spectrophotometric growth inhibition end point [98, 99].
The M27-A3 document recommends 24-h readings for caspofungin, micafungin, and anidulafungin, 24- or 48-h readings for fluconazole and amphotericin B, and 48-h read- ings for voriconazole, itraconazole, posaconazole, and flu- cytosine. An investigation of 11,000 clinical isolates of Candida species using the CLSI broth microdilution meth- odology to compare 24- and 48-h test results for fluconazole showed 93.8% categorical agreement with only 0.1% very major errors. The essential agreement, two log2 dilutions, was 99.6% [100]. The lowest categorical agreement was for C. glabrata and was predominantly the result of isolates that were susceptible at 24 h but susceptible-dose dependent at 48 h [100]. When the initial dosing and outcome data set used to establish the 48-h breakpoints for fluconazole was reevaluated using the 24-h readings, the clinical outcome and pharmacodynamic data strongly supported the early reading [101]. Again, the exceptions were those species with MIC distributions at the higher end of the range, such as C. glabrata.
Readings taken at 24 h may be more relevant for some isolates. Isolates for which the earlier reading is important show a dramatic rise in drug MIC between 24 and 48 h due to “trailing” growth. The term trailing has been used to describe the reduced but persistent growth which some iso- lates of Candida species (primarily C. albicans and C. tropi- calis) exhibit over an extended range of azole drug concentrations, making interpretation of the MIC end point difficult [102, 103]. Estimated as occurring in about 5% of isolates [104], this trailing growth can be so great as to make an isolate that appears to be susceptible after 24 h appear completely resistant at 48 h. Two independent in vivo inves- tigations of this phenomenon that employed murine models of disseminated candidiasis [103, 104] have shown that trail- ing isolates should be classed as susceptible rather than resis- tant. This concept has been corroborated by the clinical
19 Laboratory Aspects of Medical Mycology
demonstration that most episodes of oropharyngeal candidiasis due to trailing isolates respond to low doses of fluconazole, used to treat typical susceptible isolates [102].
The inclusion of a colorimetric indicator into the culture medium has been found to produce much clearer visual end points in tests with azole antifungal agents and to generate MICs that are in close agreement with those obtained using the standard broth dilution procedures [105, 106]. Commercial colorimetric microdilution plate panels for the in vitro test- ing of antifungal agents are now available for diagnostic use in the United States. Comparisons of the Sensititre YeastOne Colorimetric Antifungal Panel (Trek Diagnostics Systems Inc., Westlake, OH), which incorporates alamar blue as the colorimetric indicator, with the M27-A reference procedure have demonstrated good agreement between the methods [107–111]. In addition to clearer end points, other benefits include reduced incubation times and storage of the panels for up to 2 years at room temperature. There are also some data showing a good correlation between Sensititre YeastOne and the M38-A reference procedure [112].
Interpretive guidelines (breakpoints) have been deter- mined for fluconazole, itraconazole, voriconazole, caspo- fungin, anidulafungin, micafungin, and flucytosine against all Candida species [113–115]. Initial breakpoints for flucon- azole and itraconazole were based largely on studies of mucosal candidiasis and there were very little clinical out- come data for isolates with elevated MIC values to flucon- azole [113]. Subsequently, a reanalysis of the breakpoints incorporating additional clinical trial data and pharmacody- namic analysis, as well as consideration of mechanisms of resistance, reaffirmed the original breakpoints [116].
Breakpoints for voriconazole were developed in 2006 and were based largely on outcome data from six phase III clini- cal trials with 1,681 Candida isolates from nonneutropenic patients with candidemia [115]. The relevance of these break- points in other clinical settings remains to be established.
Interpretive criteria for echinocandin antifungal agents were published in 2008 [114]. Because of a lack of clinical isolates with elevated MIC values to the echinocandins, a single “susceptible only” breakpoint was assigned for caspo- fungin, anidulafungin, and micafungin at £2 mg/mL. A lower breakpoint would encompass almost all isolates of C. albi- cans, C. glabrata and C. tropicalis, but the higher breakpoint was chosen to avoid bisecting the distribution of C. parapsi- losis MIC values.
Following the principles established for testing Candida species and C. neoformans, the CLSI subcommittee on anti- fungal susceptibility testing has developed an approved ref- erence procedure (document M38-A) for broth microdilution susceptibility testing of conidium-forming filamentous fungi [117]. The essential features of this method include the use of a broth microdilution format, a defined test medium (RPMI-1640 broth buffered to pH 7.0 with MOPS),
an inoculum standardized by spectrophotometric reading to around 10,000 colony-forming units per milliliter, and visual determination of the MIC end point after incubation at 35°C for 24–72 h. This procedure was developed using isolates of Aspergillus species, Fusarium spp., S. apiosper- mum, and S. schenckii. The methods of inoculum prepara- tion, choice of inoculum size, time of reading, and end point selection have all been evaluated in a series of multicenter investigations [118, 119]. Nongerminated conidia are used because, at least with Aspergillus species, similar results have been obtained for germinated and nongerminated conidia [120, 121]. Quality control guidelines have been established for amphotericin B, itraconazole, voriconazole, and posaconazole [122].
Although the M27-A reference procedure served as the starting point for the M38 document, there are several sig- nificant differences between the two methods. The inoculum is about ten times higher than for yeasts and requires a dif- ferent method of preparation. Because of the differences in the size and light-scattering properties of the spores pro- duced by these fungi, the M38-A document specifies differ- ent optical densities for each genus. Careful preparation of the inoculum is essential, since a concentration outside the specified range will result in an altered MIC to most antifungal agents [123].
The end point definition is another point where the M38 and M27 procedures differ. In M27, azoles and echinocan- dins are read at a partial inhibition end point (defined as the lowest drug concentration producing a prominent reduc- tion in growth). In the M38-A2 standard it is recommended that reading the end point at 100% inhibition (no growth) better detects resistance of Aspergillus species to itracon- azole and the newer triazoles [121, 124]. For the echi- nocandins it is recommended that the minimum effective concentration (MEC) be used to define breakpoints. The MEC is the lowest concentration of drug that leads to growth of small, rounded, compact hyphal forms as com- pared to the control well.
The development of the M27 and M38 reference proce- dures for in vitro testing has provided an essential standard against which possible alternative methods can be evalu- ated. Microdilution procedures are time-consuming and labor-intensive, and there is a need for simpler and more economical methods of antifungal susceptibility testing for routine clinical use. Among the simpler methods that are now being evaluated are the agar disk diffusion tests and the Etest.
The most recent set of standards from CLSI is M44-A, a reference method for antifungal disk diffusion susceptibility testing [125]. Disk diffusion is already widely used in clini- cal laboratories for testing bacteria, so this document pro- vides a simple, rapid, and cost-effective method for susceptibility testing of fungi. Mueller-Hinton agar, which
20 M.E. Brandt et al.
may already be available in some clinical laboratories, sup- plemented with 2% glucose and methylene blue dye, is the recommended testing medium. Quality control limits for flu- conazole and voriconazole have been developed [126, 127].
Criteria have been established for fluconazole against Candida species by comparing inhibition zone diameters to MICs generated by the M27-A broth microdilution method and by testing against isolates with known resistance mecha- nisms [128, 129]. One advantage of the disk diffusion assay is that the zone diameters are read at 24 h. However, reading the prominent reduction in growth is highly subjective and can lead to susceptible isolates being categorized as resistant [130]. Otherwise, there is very good agreement between broth microdilution and disk diffusion for fluconazole [130], and even though there are no approved criteria at this writ- ing, for voriconazole as well [131]. Standards for disk diffu- sion susceptibility testing of moulds are currently under consideration [132].
The Etest (AB Biodisk, Solna, Sweden) is a patented commercial method for the quantitative determination of MICs. It is set up in a manner similar to a disk diffusion test, but the disk is replaced with a calibrated plastic strip impreg- nated with a continuous concentration gradient of the anti- microbial agent. Following incubation, the MIC is determined from the point of intersection of the growth inhibition zone with the calibrated strip. Both nonuniform growth of the fungal lawn and the presence of a trailing growth edge can make end point determination difficult. However, with expe- rience and standardized procedures, the correlation between the Etest and the M27-A reference procedure has been acceptable for most Candida species and the azole and echi- nocandin antifungal agents [111, 133–135]. For many moulds, including Aspergillus species, good correlations with amphotericin B, posaconazole and itraconazole Etest and MICs by the M38 method have been reported [136–
139]. The Etest has proved useful for the determination of amphotericin B MICs and represents one of the more reli- able ways to detect resistant isolates [140–142]. QC Etest limits for the two M27 QC isolates against amphotericin B, flucytosine, fluconazole, itraconazole, and ketoconazole have been proposed [133].
Currently, there are only three commercially available systems that are FDA approved for in vitro susceptibility testing of fluconazole. These are the Sensititre YeastOne System (Trek Diagnostics, Cleveland, OH), Etest (AB BIODISK, Solna Sweden) and the VITEK2 system (bioMerieux, Hazelwood, MO). The VITEK2 is a fully auto- mated and standardized spectrophotometric system, already in use for bacteriology in many clinical laboratories, that computes MIC values from closed cards containing dehy- drated drug. In a multicenter evaluation of 426 Candida iso- lates, the overall agreement with broth microdilution read at 24 and 48 h was 97.9% and 93.1%, respectively. Categorical
agreement with the 24-h read was 97.2% but dropped to 88.3% at 48 h. This was mainly attributable to trailing growth of C. glabrata isolates. The mean time of incubation for the VITEK2 cards was 13 h, greatly reducing the time needed for an accurate result [143]. Cards with voriconazole, flucy- tosine, and amphotericin B also show strong categorical agreement with broth microdilution [144].
The European Union Committee on Antimicrobial Susceptibility Testing (EUCAST) has developed a similar but slightly modified set of standards for susceptibility test- ing of yeasts that result in essentially the same MIC values as the CLSI methodology [145, 146]. Quality control strains and quality control ranges for amphotericin B, flucytosine, fluconazole, itraconazole, posaconazole, and voriconazole have been established [147]. The EUCAST methodology has some notable differences from the CLSI method. Both use RPMI 1640 as the growth medium, but the EUCAST methodology adjusts the glucose concentration to 2%, which has been shown to enhance the growth of most yeasts without significantly impacting the MIC value [148]. The inoculum used in the EUCAST methodology is 50 times that dictated by CLSI. Finally, all MIC values are deter- mined by measuring absorbance using a plate reader, and they are determined after 24 h of incubation [145, 149]. For these reasons, the CLSI breakpoints cannot be applied to isolates whose MIC values were determined by the EUCAST methodology [150].
For fluconazole, the MICs generated by the EUCAST method and by the CLSI method are essentially the same up to 2 mg/mL, but above this level the CLSI generated MIC values are twofold higher [150]. The result is that there are two sets of breakpoints for fluconazole depending upon the standards on which the testing was based [113, 145]. It must be borne in mind that the EUCAST breakpoints currently only include the species C. albicans, C. tropicalis, and C.
parapsilosis, do not include a susceptible dose-dependent category, and the pharmacodynamic analyses used to evalu- ate the breakpoints were based on doses of 400 and 800 mg [145]. That being said, when 475 Candida isolates were tested by the two methodologies simultaneously and the respective breakpoints were applied, there were only eight major discrepancies. Three isolates that were resistant by the EUCAST method were susceptible by the CLSI method, and five isolates that were susceptible by the EUCAST method were resistant by the CLSI method [150]. The latter five isolates all had a trailing growth phenotype and all were susceptible by the CLSI methodology when MIC values were determined at 24 h. Breakpoints for voriconazole have been established for C. albicans, C. tropicalis, and C. parap- silosis. Like the fluconazole breakpoints, the EUCAST vori- conazole breakpoints are much lower than the CLSI breakpoints and there is no susceptible dose-dependent category [151, 152].